You have no idea what AI is doing to software

**[00:00:00]** AI and what it means for software. This is going to be fun. I can't believe we're actually doing this stream. By the way, there's this is a little meta, but the let me know, of course, if you if you hear me. Looks good. And there it is in LinkedIn. Yes, perfect. And I think it's going on X as well. But let me know in the comments in any of these platforms if you actually see me and everything is looking good. The reason I'm saying that is I am multistreaming from my computer of course, but I am streaming from a tool that we just built two days ago. So, this is a brand new podcasters dream tool that allows us to have guests on the show, have collaboration on the show, have multi-hosts, have app access, mobile phone access where you can join, but also control the actual stream and a bunch of other things. And we can talk about that later uh if if anyone's interested. But that app I literally built in 24 hours and and it's what we're running right now where you're seeing me right now and it's running multiple platforms multistreaming. Can't believe it's actually working. So I'm on here in my computer. It's actually running back here on my OpenClaw if you know what that is. And we can talk about that as well. And it's really fun because I could do the stream here. I have my horizontal, vertical, and then I'm streaming over to Instagram. I'm streaming over to YouTube. I'm streaming over to X. I'm streaming over to LinkedIn. All at the same time from an app that we just built in 24 hours, which to me is mind-blowing. There is so much to unpack here. So, AI, **[00:02:00]** we've all seen it. We've all experienced it. uh you have your Chad GPTs of the world, your anthropics, your Geminis, and you might use it to ask for recipes and you might use it for some work related tasks. Doesn't really matter what you're using for you. You probably interacted with it a little bit. In the software development side of the world space, AI has brought with it a big paradigm shift in how we work with applications, how we work with software. And that's what I want to unpack today. I want to talk about what it means for software developers. By no means it's software developers don't have a job anymore. On the contrary, it means one software developer can now do what a team of six or seven software engineers could do before. It is outstanding. So, I'll give some examples during the chat here. If you have any questions, please drop them below. And if you like the content, please like and subscribe in whatever platform you're using. Uh, if you're here on YouTube and make make sure you ring the bell. If you are on X in LinkedIn, Instagram, you can follow as well or connect if you want. Also, I'll see if I can keep open the chats of the different platforms. Not all of them have a little chat feature. So, I can see on LinkedIn, I'll be swapping through the tabs in case somebody has any questions on I don't know. I don't know if an X I can actually see the comments. I think I I would have to go to the stream itself. So we can try that and see what that looks like on that end. But on Instagram, I will see the comment feed as well. And then on YouTube also. So what has changed and why is everyone going nuts in the past few days about **[00:04:00]** AI in general, but more specifically some of the tools that have come out. And I was speaking with Mike Carlo about this the other day on the stream what OpenClaw clawbot and all these variation name variations of the same thing mean and I'll I'll tell you a little bit about it and I'll give you an example of why I think this is so groundbreaking. Some people use it as an assistant, a glorified assistant. It's great because you can have a tool now that you can chat with via WhatsApp, via Telegram, via Slack, via Teams, wherever you have your conversations. You can actually interact with these tools and it's really nice. But what is it that makes me excited about all this is not only that you can chat with it from wherever you are which is a fantastic plus and I'll get into that as well but the fact that it can do things proactively. So if you have used chat GPT in the past typically you open it in your computer, you open it in the browser, you open it in the app and you ask it a question, it gives you an answer. You ask it to do research, it'll go, it might do some thinking, it might spawn some additional tasks and then it will come back and uh give you a response. But that's basically it. So there is a back and forth mechanism. You say something, it comes back. you might send it off to do extended research and yes it will notify when it's done but that's pretty much the extent of it. Some of these tools have advanced and progressed and now they have a way to within the app itself have the sort of that coding aspect where it can it can actually do more things with files that for developers has existed for a while now through their CLI Claude Code for instance from anthropic really useful because if you're a developer you're already using a tool like VS code or cursor or **[00:06:00]** whatever you're using and There's you can then code there and have this this side friend that allows you to interact with your codebase and ask it questions and have it code things for you, etc. But again, it's a back and forth kind of thing where you tell it, hey, would can you please make a page that does this? It'll go do it and tell you, I'm done. But it'll stop. That's the important thing. Then you can say, "Oh, well, in the more advanced models, you can say, well, I have this plan. Let's execute this plan and go ahead, please do it." And so it might do one or two steps from it and then stop and it'll say, "I'm done." And that's it. That's as far as it goes. So very powerful model. It understands you very well. It knows programming perfectly, but it doesn't have free reign on continuing to to do things. So that's a limitation, right? And still not complaining, amazing tools, amazing platforms. We have just been tearing through bug fixes and things like that just because of the ease of use that these tools provide. So before I get into all the OpenClaw and cloudbot and all that good stuff, I want to talk about just as a regular like just how powerful is an LLM for a software developer or anyone who has a software that they're trying to debug or make better etc. Let me give you some examples. So before what would happen is we would get a ticket from a client or someone on behalf of a client and they would say hey this thing is not working as expected this particular edge case is showing up maybe uh we were talking about a few days ago there's transcript generated for an LMS and perhaps for **[00:08:00]** that particular person not everyone but that particular person the transcript is not being generated. So before we would have to see okay so who is this person let me find out and we would have to go and see all their data and then compare them against other people and see okay so this person generated this person did not generate. So we try to find similarities and differences until we can determine well we have an example of these fields and these fields and they both work. Why is this one not working? So oh okay this particular field might look different. Then we would go into the codebase and look at that function and see what happens when this field is set this way. So we go through all the conditions and we sort of have to calculate it in our minds and follow through and see oh okay so it looks like maybe there's something missing here. And that would take us a few minutes maybe a few hours depending on the complexity of the database and the program and what things it does because maybe it's five different functions that are running and it's a little flow. So in those cases we would add debugging. We would add some code in it in a staging server of course where we could try to generate the transcript and then it would go through and say okay a bunch of logs in a terminal it would show us like this passed this passed this passed didn't this didn't pass right that there's the issue okay so then we would go in yeah now that we see the issue because of the log that we created the debug logs now we can go in and modify the code or figure out what's happening first modify the code and then run it again and fix it. So again, that would take from 10 minutes to an hour or multiple hours depending on the complexity of the codebase or of the project. And it's okay. I mean, that's that's what we do. And it's it's been a fun experience sometimes debugging. Not everyone will agree, but it's it's a it's an adventure with LLMs, specifically with Claude Code, which is my tool of choice at this moment, and that can change at any moment. **[00:10:00]** It's it is a fun experience. Like it literally becomes a fun experience. Why? Because it has access to the whole codebase. What does this mean? And I think I'm going to move my display here because of course our new application is optimized to any screen. So I can put on a teleprompter and I can see everything and control everything the same way. So fun. Okay. So what happens now is I'll open VS code I'll open whatever tool you're using and if I get that same request then what will happen is I will grab that request copy paste it into the LLM and tell claude this is happening for this person these are the fields or or even this is the ID of the person and then Claude can go through check all the relevant files because he already has an understanding of how that system in particular works. Find the issue. Figure out what in the code is different. Find out why for them it's not working. Figure out what the solution should be. Make a fix and test it all in a matter of seconds. So you can imagine the amount of time and money you save with these tools because now of course it's not like yeah just believe it blindly. We're talking about software engineers here. It's it's a you have a tool that allows you to as far as you let it give you the solution immediately. So I can tell it tell me what's happening and it'll just tell me and that's it. Or I can say, "Tell me what happened and tell me how to fix it." And it will go, "Tell me how to fix it. Show me the diff. Show me the two codes side by side. Show me where it changes and what changes." And then I can decide, "Yeah, that looks good. That's how I would fix it if I saw that same issue happen." And then you can say, "Fix it." Or I can just grab **[00:12:00]** the code and put it in myself. But now I know what happened in seconds. So you can imagine if you're managing a production system that has thousands or hundreds of thousands of users and something's going wrong for many of them and you can push out a fix in less than a minute. I mean that's unheard of. That wasn't a possibility before unless you had like the even if you had the best developer in the world or software engineer. Yes. Yes. There are people who are very good at debugging and fixing things in real time and that's amazing. But having that power at the disposal of everyone, every company, it's incredible. I'm I'm mind blown. So, we've been working like this for months, which means we've been saving time when fixing things or when adding features. And that also means lower costs for our clients. So if you have some sort of agreement or retainer or something with a client and you now can fix something extremely fast then that's less money that they have to pay and you can manage it the way you want right of course for certain projects you'll you'll keep a balance such that you're still you should always by the way if you're software engineer you shouldn't be doing hourly you should be doing by project and so whatever your project provides as value to the customer that's what it's worth it's not your time right especially with these tools you're going to do things much faster So time I mean doesn't make sense. But if you do have a support retainer of a sorts where you're assisting, you're helping, you are providing support services, then amazing because you save time and you save them money. It's a win-win right there. Um can handle it as you will. But but I think I mean I don't know. I like to be in the position of my clients as well as if I were someone else's client. And um if something can be done more efficiently, fantastic. Right now, **[00:14:00]** projectwise, this means a software company can now push out a lot more software a lot quicker and provide solutions that are more tailored to the user. And we talked about this in another show where it's now the era of software that's aligned to you and not you having to align yourself to software. At least that's what we think because when before you might have to fit yourself into an Asana or a Jira or an SAP, Oracle, whatever system you sort of think that might work for you or your company, you have to fit into their box. And now these tools allow us to make our own tools. Our own tools pretty much in real time and use them the way that we need to use them. That's what's so special about it that now we can actually build things that we need to use and that we're going to use because it fits into our workflow. And what's even better is that you can create an MVP, minimum viable product. Hey, this works. I use it. Hey, it doesn't work. Throw it away. Took five minutes. I don't care. So, it's like there's there's nothing like that has ever ever existed before. And now with the coming of OpenClaw, clawbot, moldbot, whatever you want to call it, it is truly remarkable what you can do. I'll give you some examples as well that actually let me give you one real example. So this whole application that I'm using right now for streaming for so through which I am streaming **[00:16:00]** to one two three four different platforms at the same time brand new tool and I can add effects and I can put images on top. I can share screen. We we'll try we'll try share screen and that's a funny story too. Um and then I can bring in guests. So if anyone wants to join you can just let me know. I'll send you a link. So I have it'll generate links for guests that just appear here in the bottom and you can say hi and stuff or it has collaborators that can join in just be part of the call and then also it has um a way to multihost so you can have multiple hosts at the same time but it's it's brilliant. It just really works. Now what I really like about it is that when you are building something typically you you wait or you stop yourself because you have bias and you have limiting beliefs. So you see now I'm never going to build that. I mean it's going to take years or months or whatever it's going to I don't know I have to focus on projects. I I can't focus on just creating a tool for myself. Well, now you can. Like this tool was built in 24 hours. And here we are streaming. And I Let me show you what. So, what's really funny is that you just tell it things and make adjustments to it. And I was showing my wife yesterday. We were working on Mom's Guide essay and I see we'll test the live stream. I mean, the sharing so we can see if it's actually true. But we were looking at it and we're like, "Oh, let's let's add this new little thing that we want to add." and I just told it and it just added it. So now you have this possibility of having like an employee which is an AI agent that you can talk to via messages or voice whatever you want and it will listen to you and then it will do the things. So it will actually modify the applications. What I want to show you is a feature that we added in a second, right? I want to just show you the **[00:18:00]** feature itself and then I'm going to show you another platform that I built yesterday just just for kicks for the podcast which you should visit webtalk.show by all means. Um you can see what uh what that looks like and if you haven't heard our recent episodes on mult open etc. You can find those here and you can even do chat I mean search and just find everything in everything we've ever said on the podcast you can find immediately and I'll talk about that a little bit too because I want to do a demo on on that part. It's again ridiculously good. So first I want to show you for those who like the whole podcasting thing and and streaming etc. This tool that I'm using right here that you can see it here on my mobile device as well if it decides to focus. Those are the shows, right? And then I can click enter studio. And if I click that, I just don't want to click it before I give you a heads up. I will join this call as a host immediately. Immediately. Like literally immediately. Not like Zoom takes a little bit for No, no, no. It just it just jumps in and then I can control the the call or whatever I want. Or I can just jump. It's brilliant because compared to other tools that are good, we use Riverside for regular streaming and things like that, but we're going to be switching to this, of course. It's you're sort of limited. You're in that room and if you leave a room or you whatever, it it sort of dies. It's not it's not necessarily persistent. And the whole experience for guests, it again, not not the best, but it's nice what for what exists up to this day. It's nice. That's what we use. But this, you just get a link, you click it, immediately in your browser, you're in the call. And it's amazing because now I could, if I wanted to, join here, then walk out, turn off this one, and get in a car and continue the stream in the car from my phone, and then get somewhere else, and then join again from the computer. And now we're back. Or I could use this as multiple cameras, multiple angles. I can turn on multiple if I have multiple computers. I don't **[00:20:00]** have multiple cameras, but I have multiple computers. Let's say you have multiple computers and multiple cameras, but you don't have a mixer or you don't have inputs that you can put in multiple or your computer doesn't support like you cannot handle for 4K for many computers at the same time or or many cameras in one computer. Well, you connect four computers and then each one can just connect to the call and now you have your multi-angle setup right here uh in the show as well. We're going to add some features to it so that you can like actually do live production um kind of deal in it. So that's going to be super fun as well. But let me join and just to see if it's still working. I don't know. It's new. So if I do enter studio, it's going to say allow microphone and camera. So here I am. And then let me just mute myself because this is not going to be practical if I'm Oh my goodness, I left. There it is. Okay. Close. There we go. Okay. So, here I am, right? And volume. So, here I am on my phone. I just I just joined directly. I can have the sidebar. You don't see it here, but there it is. So, that's that's what it looks like where it's it's showing the whole production toolkit that we built into the thing. I don't know why it still has volume and it's how can this how can this have bu been built in 24 hours? Like it's not it doesn't make any sense. Anyway, I'm going to leave this one. But now you see here I am right. So I'm going to stop that. It is it is incredible. It's the fact that LLMs and technology allow us to build tools exactly for what we need them. That is where the real power lies. Super cool if it worked. **[00:22:00]** All right. Yeah. Let me know if you hear me. That would be that would be super cool. Okay. So, I'm streaming again. I I changed something while we were streaming. So, I changed something in the application itself. And so it's the connection to the stream was reset, but then the app itself didn't know that that had stopped. So I I just saw the stop button. So I I made the adjustment for that, but I wanted to leave it on because that way it would allow us to continue in the stream. So I don't know if that's actually what's happening here. We'll see. I want to see if it actually brings in the stream. Super cool. All right. So, I will start another stream so that this one is has it doesn't have a huge break there, but I'll do it if I have a moment right now, if not later. But I want to show the page that we made for the Web Talk Show so that you can see what in action what it means to have something where you can tell the tool to actually do changes. So we'll do that. But I'll jump out before. All right, there we go. Yeah, it came back to life. I can't believe it. Anyway, talk soon. Cheers.